SVD workshop in Procida

New!!!: How to get to Procida

Ferries (traghetti) will leave from Calata Porta di Massa (C on the map); hydrofoils (aliscafi) will leave from `molo Beverello' (B on the map) . See below: the map from the conference site to the molo Beverello and to Calata Porta di Massa, and the timetable (with highlighted the suitable departure times). We look forward to see you in Procida!

Timetables for ferries and hydrofoils

CAREMAR ferry from molo Beverello (Calata Porta di Massa ) - 10:45;
SNAV hydrofoil from molo Beverello - 12:30 (approx 40' ride);

If you decide to take the ferry, there is a free shuttle to go from molo Beverello to Calata Porta di Massa.

Lunch will take place at 13:30 at ristorante CRESCENZO

The workshop will start at 15:00, as indicated in the schedule

Dinner will take place at 20:30 at ristorante La Pergola

In the map you will find the directions from the Procida harbour (A), to the Crescenzo restaurant (B), to the conference place (C) and to La Pergola (D). It takes a 25' walk from (A) to (B) and a bit more from (B) to (C); a taxi service is available (price 15 euros for four people, each way).

Workshop on Singular Value Decomposition (SVD) after CARME, with guest speaker Trevor Hastie

As an extra event of CARME this year, a one-day (24th afternoon and 25th morning) workshop on the theme of the singular value decomposition (SVD) has been organized on beautiful Procida Island, close to Naples. The SVD is by far the most useful mathematical result underlying CARME, and finds an increasing number of applications in modern data science. The workshop consists of four sessions of 1.5-2 hours each, starting with an overview of the SVD, its various interpretations and classical applications (Michael Greenacre), then the modern developments of sparse SVD, matrix completion, subspace averaging, page rank and nuclear-norm regularization (Trevor Hastie), computation of the SVD for large data sets and streaming data (Alfonso Iodice d'Enza and Angelos Markos) and finally the generalizations of the classic SVD to three- and multi-way data (Pieter Kroonenberg). The event requires a separate registration, which is cheaper for CARME participants.